Binary-Origami Figurator

Infrastructure for pragmatic epistemology. Combining

• i-docs navigation,

• PROMPT epistemological scoring, and

• boundary objects theory.

An evidence graph for investigative journalism.

Status: Phase 1 (PoC) - v1.0.0 Release Version: 1.0.0

New Here? Start with the Wiki

Tip	Confused by the terminology? The Binary-Origami Wiki explains everything in metaphors, diagrams, and plain language.

This isn’t just a database. It’s infrastructure for folding and unfolding evidence—so everyone can see the shape that fits their needs.

Why "Binary-Origami"?

• Binary: Evidence is stored as clear, connected data (supports/contradicts, 0-100 scores)

• Origami: The same evidence can be "folded" into different forms for different audiences

• Figuration: The rules for folding/unfolding are transparent and reversible

Concept	Description	Learn More
i-docs Navigation	"Choose Your Own Adventure" for evidence	Wiki
PROMPT Scoring	"Nutrition labels" for trustworthiness	Wiki
Boundary Objects	"Shared maps" with multiple routes	Wiki
Evidence Graphs	The "skeleton" beneath the origami	Wiki

For the Impatient

1. Try the Quick Start to see it in action

2. Skim the Binary-Origami Metaphor page

3. Dive into the FAQ if something’s unclear

Vision

We didn’t fall from Truth to Post-Truth; we evolved to complex epistemology without building infrastructure. This system IS that infrastructure.

Core Concepts

1. i-docs Navigation: Navigation over narration, reader agency

2. PROMPT Framework: 6-dimensional epistemological scoring (Provenance, Replicability, Objective, Methodology, Publication, Transparency)

3. Boundary Objects: Multiple audience perspectives on same evidence

4. Evidence Graph for Investigative Journalism amd Related Disciplines

Quick Start

Prerequisites

• Elixir 1.18+ & Erlang/OTP 27+

• Phoenix 1.8+

• Podman & podman-compose

• just (task runner): https://github.com/casey/just

1. Clone Repository

git clone https://github.com/Hyperpolymath/bofig.git
cd bofig

2. Start Databases

podman-compose up -d

Verify ArangoDB is running: http://localhost:8529 (root/dev)

3. Full Setup (deps, databases, seeds)

just setup

Or manually:

mix deps.get
mix ecto.create
mix run -e "EvidenceGraph.ArangoDB.setup_database()"
mix run priv/repo/seeds.exs

4. Start Phoenix Server

just dev

Visit: - Application: http://localhost:4000 (register/login required) - GraphQL Playground: http://localhost:4000/api/graphiql (dev only) - Health Check: http://localhost:4000/api/health

GraphQL API Examples

Query: Get All Claims

query {
claims(investigationId: "uk_inflation_2023") {
idtextclaimTypeconfidenceLevelpromptScores {
provenancereplicabilityobjectivemethodologypublicationtransparencyoverall
}
supportingEvidence {
evidence {
titleevidenceType
}
weightconfidence
}
}
}

Query: Evidence Chain (Graph Traversal)

query {
evidenceChain(claimId: "claim_1", maxDepth: 3) {
rootClaim {
text
}
nodes {
...onClaim {
idtext
}
...onEvidence {
idtitle
}
}
edges {
relationshipTypeweightconfidence
}
maxDepth
}
}

Mutation: Create Claim

mutation {
createClaim(input: {
investigationId: "uk_inflation_2023"text: "Inflation disproportionately affected renters"claimType: SUPPORTINGconfidenceLevel: 0.85promptScores: {
provenance: 70replicability: 65objective: 75methodology: 70publication: 65transparency: 70
}
}) {
idtextpromptScores {
overall
}
}
}

Mutation: Import from Zotero

mutation {
importFromZotero(
investigationId: "uk_inflation_2023"zoteroJson: {
key: "ABC123"itemType: "journalArticle"title: "New Economic Study"url: "https://doi.org/10.1111/example"creators: [{name: "Smith, J."}]
tags: [{tag: "economics"}]
}
) {
idtitlezoteroKey
}
}

Query: Navigation Paths

query {
navigationPaths(
investigationId: "uk_inflation_2023"audienceType: RESEARCHER
) {
idnamedescriptionpathNodes {
entityIdentityTypeordercontext
}
}
}

Mutation: Auto-Generate Navigation Path

mutation {
autoGeneratePath(
investigationId: "uk_inflation_2023"audienceType: SKEPTIC
) {
idnamepathNodes {
entityIdorder
}
}
}

Project Structure

bofig/
├── lib/
│ ├── evidence_graph/ # Core business logic
│ │ ├── claims/ # Claims context
│ │ │ └── claim.ex
│ │ ├── evidence/ # Evidence context
│ │ │ └── evidence.ex
│ │ ├── relationships/ # Graph edges
│ │ │ └── relationship.ex
│ │ ├── navigation/ # Audience paths
│ │ │ └── path.ex
│ │ ├── arango.ex # ArangoDB client
│ │ ├── prompt_scores.ex # PROMPT scoring
│ │ └── application.ex # OTP supervisor
│ └── evidence_graph_web/ # Phoenix web layer
│ ├── schema/ # GraphQL schema
│ │ ├── types/ # Type definitions
│ │ └── schema.ex # Root schema
│ ├── endpoint.ex
│ └── router.ex
├── priv/repo/
│ └── seeds.exs # UK Inflation 2023 test data
├── config/ # Environment configs
├── docs/ # Architecture docs
│ ├── database-evaluation.md
│ └── zotero-integration.md
├── ARCHITECTURE.md # Data model, API design
├── ROADMAP.md # 18-month plan
├── CLAUDE.md # AI assistant context
├── Containerfile # OCI container build
└── podman-compose.yml # Container orchestration

UK Inflation 2023 Test Dataset

The seed data includes a complete investigation:

• 7 Claims (primary, supporting, counter)

• 10 Evidence items (expand to 30)

• Official statistics: ONS CPI, Ofgem, BoE

• Academic: Peer-reviewed studies

• Think tanks: Resolution Foundation, IFS

• Interviews: Expert opinions

• 10 Relationships (supports/contradicts/contextualizes)

• 3 Navigation Paths:

  1. Researcher: Evidence-first, methodology priority

  2. Policymaker: Authoritative sources, recommendations

  3. Affected Person: Personal impact, clarity

PROMPT Score Examples

| Evidence | Prov | Repl | Obj | Meth | Pub | Trans | Overall | |----------|------|------|-----|------|-----|-------|---------| | ONS CPI Data | 100 | 100 | 95 | 95 | 100 | 95 | 97.5 | | Academic Study | 85 | 80 | 75 | 85 | 90 | 75 | 81.8 | | Think Tank Report | 75 | 70 | 65 | 75 | 80 | 70 | 72.3 | | Expert Interview | 85 | 45 | 60 | 50 | 40 | 75 | 59.0 |

Development

Run Tests

mix test

Interactive Shell

iex -S mix phx.server
= Query ArangoDB directly
iex> EvidenceGraph.ArangoDB.query("FOR c IN claims RETURN c")
= Get a claim
iex> EvidenceGraph.Claims.get_claim("claim_1")
= Evidence chain traversal
iex> EvidenceGraph.Relationships.evidence_chain("claim_1", 3)

Code Quality

= Format code
mix format
= Static analysis
mix credo
= Type checking
mix dialyzer

Deployment (Phase 2)

• Hosting: Hetzner Cloud (EU data sovereignty)

• ArangoDB: ArangoDB Oasis (€45/month)

• Phoenix: Systemd service, Nginx reverse proxy

• CI/CD: GitHub Actions

Documentation

Conceptual (Start Here!)

• Binary-Origami Wiki - Explains the metaphor and concepts in plain language

• The Metaphor - Why "Binary-Origami Figuration"?

• Folding 101 - Hands-on tutorial

• FAQ - Common questions answered

• Glossary - Term definitions

Technical

• ARCHITECTURE.md - Data model, database design, API specs

• ROADMAP.md - 18-month implementation plan

• docs/database-evaluation.md - ArangoDB comparison

• docs/zotero-integration.md - Two-way sync design

• CLAUDE.md - AI assistant context

Key Features

Implemented (v1.0.0)

• Multi-model ArangoDB integration (document + graph)

• GraphQL API with Absinthe (15 queries, 11 mutations)

• PROMPT epistemological scoring (6 dimensions, audience weighting)

• Claims, Evidence, Relationships, Navigation Paths data models

• Graph traversal algorithms (evidence chains, shortest path, contradiction detection)

• Zotero REST API (import, export, batch-import, sync-status)

• Phoenix 1.8 LiveView frontend (5 pages: Dashboard, Investigation, Graph, PROMPT, Navigation)

• User authentication (phx.gen.auth with bcrypt, magic links)

• D3.js force-directed graph + radar chart visualisations

• Audience-weighted navigation paths (6 types)

• UK Inflation 2023 test dataset (7 claims, 30 evidence, 38 relationships)

• Production deployment (Containerfile, nginx, systemd)

• NUJ user testing protocols

• A2ML v2.1 Cyberwar-Ready Trustfile

• 257 tests, 0 failures, full RSR compliance

Coming Next (Phase 2)

• Zotero browser extension (one-click import)

• Multi-investigation dashboard

• Real-time collaborative editing

• Advanced visualisations (timeline, heatmap, Sankey)

• IPFS provenance integration

• Hetzner Cloud deployment

Philosophy

This isn’t just a database. It’s infrastructure for coordinating without consensus.

Every design choice asks: 1. Does this support multiple audience perspectives? 2. Does this make epistemology measurable? 3. Does this enable navigation over narration?

Contributing

Open source from day 1. See [ROADMAP.md](ROADMAP.md) for planned features.

Month 3 = Decision Point: User testing with 25 NUJ journalists determines go/no-go.

Related Projects

• LithoglyphDB - The narrative-first, reversible, audit-grade database

• GPNL - Dependently-typed Glyph Projection Language (compile-time proofs)

• Lithoglyph Studio - Zero-friction GUI for non-technical users

• Zotero-Lithoglyph - Reference manager with PROMPT scores

License

MPL-2.0 (Palimpsest License)

Contact

• Repository: https://github.com/Hyperpolymath/bofig

• Issues: https://github.com/Hyperpolymath/bofig/issues

• User Testing: NUJ network (Month 3, 6, 12)

---

Built with: Elixir, Phoenix, ArangoDB, Absinthe, LiveView, D3.js

Inspired by: i-docs (PMPL-1.0 Open Doc Lab), Boundary Objects (Star & Griesemer), Pragmatic Epistemology

Last Updated: 2026-02-21

Architecture

See TOPOLOGY.md for a visual architecture map and completion dashboard.